上周跟一个做投资的朋友吃饭,他说了句话让我直接宕机了:
"我每年要看好几十个赛道,每个都要写报告。但说实话,前两周基本全浪费在找资料上,真正思考的时间没几天。"
我问:那你没用 AI 吗?
他说:用了啊,让 AI 搜一下行业情况,回答都是片汤话,最后还是得自己查。
你看问题就出在这儿了。大部分人用 AI 做行业研究,是把 AI 当搜索引擎用了。问一句答一句,信息还是碎的。
前两天我刷到一篇文章,讲怎么用 Codex 快速拆解一个陌生行业。作者的方法论让我豁然开朗:不是让 AI 搜信息,而是让 AI 帮你搭建一整套行业的操作系统。
读完我立刻用 Workbuddy 试了一遍,发现不仅可行,而且最终还能把整个流程打包成一个 Skill,以后换行业,一句话就搞定。
下面我拿"宠物冻干食品"这个行业当例子,一步步带你操作。每一步都有具体的指令,你照着来就行。
第一步:先建数据库,别用人脑记
大部分人研究新行业,是看到什么就记什么,今天刷到一篇文章、明天收藏一个视频。信息是散的,脑子是乱的,过两天全忘了。
正确做法是:打开 Workbuddy,直接说:
"帮我做一个宠物冻干食品行业数据库。搜索行业信息,整理品牌、产品、用户痛点、关键词、内容渠道。输出 Markdown 格式。"
Workbuddy 会自己去搜集信息,几分钟后吐出类似这样的目录结构:

这一步做完,你对这个行业就有了一套基本信息盘。不用记,随时翻。
第二步:解剖一个竞品,看清它怎么赚钱
信息有了,接下来弄明白:这个行业到底靠什么赚钱?
随便挑一个品牌,比如"朗诺",打开它的天猫旗舰店,然后跟 Workbuddy 说:
"分析朗诺的天猫旗舰店页面结构,告诉我它的产品线怎么分的、价格怎么定的、详情页强调哪些卖点、消费者评价里高频提到哪些词。"
Workbuddy 会把页面从头到脚拆开:
-
产品线:6大品类30+系列
-
定价:¥9.9试吃 → ¥29-39小包装 → ¥111-131大包装
-
详情页卖点:第一层品牌背书、第二层工艺差异区、第三层产品展示
-
评价高频词:回购、好吃、挑食改善,但最大的短板是客服

看到没有?别人的卖点和坑点,一次性全暴露了。 你就算零基础,也立刻知道了"棒棒糖/雪糕/耳袋"这套社交行话。
第三步:扒内容生态,找到流量密码
行业怎么赚钱搞清楚了,接下来搞清楚:他们到底靠什么获客?
让 Workbuddy 做这件事:
"整理宠物冻干食品行业在抖音和小红书上粉丝最多的 30 个账号,分析最近三个月什么类型的内容数据最好。"

结果出来后,进一步追问:
"把这些内容按功能分类,哪些是拉新的,哪些是收藏型的,哪些是直接带货的。"
最终你会得到一张规律表:

到这你就明白了:这个行业的内容是有套路的,不是随便发就能火。 拉新型制造焦虑,收藏型提供价值,带货型用真实体验触动消费。
第四步:搭知识地图,把点连成面
前三步都是散点,第四步把它们串成一张网。
跟 Workbuddy 说:
"帮我画一张宠物冻干食品行业的知识地图,用树状结构展示一级、二级、三级分类。每个节点列出关键公司和最近一年的趋势。"
它输出:

有了这张地图,行业结构不再是脑子里模糊的抽象概念,而是实实在在地摆在你面前。哪块能进、哪块别碰,一目了然。
第五步:打包成 Skill,以后一键复用
前面四步走完,你已经用差不多一小时把这行业摸得七七八八了。
但真正的重点是:下次换行业,你不想再重新教 Workbuddy 一遍。
把上面这整套流程打包成一个 Skill。以后只需要说:

"用快速了解行业的 Skill 分析一下智能家居行业"

注意:很多小伙伴都说创建完了无法使用,你要在「技能」栏选中skill才能发挥作用。
Workbuddy 自动走完建库、拆竞品、挖内容、搭地图一整套。一小时变一句话,这才是这套方法真正的价值。
你甚至可以加上自动更新机制,设置行业信息源、监控竞品变化、每周自动生成简报。你的知识库从死的变成了活的。
写在最后
饭桌上那位做投资的朋友后来也试了一遍。他跟我说了一句话:
"以前麦肯锡做行业调研最少两周起步。但现在,靠自己一个人加上 Workbuddy,一小时也能摸到八九成的信息。"
说到底,这套方法不是在教你研究一个行业,而是在教你如何把学习行业这件事,变成一套流程、一个工具、一个开关。
(图中数据仅作为参考使用,并不代表实际数据)
我是喜乐,正在用 AI 搭建一人公司,记录从 0 到 1 全过程。
如果你也想一起长期学习AI,不想被时代甩下车,我建了一个学习群,私聊我。适合想学AI、想做AI副业、不想被时代淘汰的人。
Last week I had dinner with a friend who works in investing, and one thing he said made my brain just crash:
"Every year I look at dozens of sectors, and I have to write a report on each one. But honestly, the first two weeks are almost entirely wasted on digging up materials. The actual thinking time only takes a few days."
I asked: So aren't you using AI?
He said: I do use it. I ask AI to look up the industry situation, but the answers are all bland generic filler. In the end I still have to check things myself.
See, that's exactly where the problem lies. Most people use AI for industry research by treating AI like a search engine. Ask one question, get one answer, and the information stays fragmented.
A couple days ago I came across an article about how to use Codex to quickly break down an unfamiliar industry. The author's methodology suddenly made everything click: instead of asking AI to search for information, you get AI to help you build an entire operating system for the industry.
After reading it, I immediately tried it with Workbuddy and found that not only is it feasible, but you can even package the whole process into a Skill. Next time you switch industries, one sentence gets it done.
Below I'll use the "pet freeze-dried food" industry as an example and walk you through it step by step. Every step has specific instructions, so just follow along.
Step 1: Build a database first — don't rely on your brain to remember things
Most people researching a new industry just jot down whatever they see. One day they stumble on an article, the next day they bookmark a video. The information is scattered, the mind is a mess, and within two days they've forgotten all of it.
The right approach is: open Workbuddy and simply say:
"Help me create a database for the pet freeze-dried food industry. Search for industry information, organize brands, products, user pain points, keywords, and content channels. Output in Markdown format."
Workbuddy will go collect the information on its own, and within a few minutes it spits out a directory structure like this:
Once this step is done, you have a basic information base for the industry. No need to memorize anything — you can flip through it anytime.
Step 2: Dissect a competitor and see how it makes money
Now that you have the information, the next thing to figure out is: what exactly does this industry rely on to make money?
Just pick any brand, like Langnuo, open its Tmall flagship store, and say to Workbuddy:
"Analyze the page structure of Langnuo's Tmall flagship store, tell me how its product lines are divided, how the prices are set, what selling points are emphasized on the detail pages, and which words come up most frequently in consumer reviews."
Workbuddy will tear the page apart from top to bottom:
Product lines: 6 major categories, 30+ series
Pricing: ¥9.9 trial pack → ¥29-39 small packs → ¥111-131 large packs
Detail page selling points: first layer is brand endorsement, second layer is the process differentiation zone, third layer is product showcase
High-frequency review words: repurchase, tasty, picky-eater improved — but the biggest weakness is customer service
See that? Other people's selling points and pain points are all exposed at once. Even if you start from zero, you instantly pick up the social media jargon of "lollipops / ice cream / ear bags."
Step 3: Dig into the content ecosystem and find the traffic formula
Now that you understand how the industry makes money, the next thing to figure out is: how do they actually acquire customers?
Have Workbuddy do this:
"Compile the 30 accounts with the most followers in the pet freeze-dried food industry on Douyin and Xiaohongshu, and analyze what types of content performed best in the last three months."
Once the results come out, push further:
"Classify this content by function — which is for new customer acquisition, which is for saving/bookmarking, and which directly drives sales."
In the end you'll get a pattern table:
At this point you understand: content in this industry follows a playbook, and it's not something you can just post randomly and expect to blow up. The acquisition type manufactures anxiety, the bookmark type provides value, and the sales type uses real experiences to trigger purchases.
Step 4: Build a knowledge map and connect the dots into a bigger picture
The first three steps gave you scattered points; step four weaves them into a network.
Say to Workbuddy:
"Draw me a knowledge map of the pet freeze-dried food industry, using a tree structure to show the first, second, and third levels of classification. List the key companies and trends from the past year for each node."
It outputs:
With this map, the industry structure is no longer a vague abstract concept floating in your head — it's right there in front of you. Which areas you can enter and which ones to avoid become clear at a glance.
Step 5: Package it into a Skill and reuse it with one click later
After the first four steps, you've already gotten a solid grasp of the industry in about an hour.
But the real key point is: next time you switch industries, you don't want to have to teach Workbuddy all over again.
Package the entire process above into a Skill. From then on, you only need to say:
"Use the 'quick industry overview' Skill to analyze the smart home industry"
Note: many people say they can't use it after creating it — you need to select the Skill in the "Skills" panel for it to work.
Workbuddy automatically runs the whole process: building the database, dissecting competitors, mining content, and building the map. One hour turns into one sentence — that's the real value of this method.
You can even add an auto-update mechanism, set up industry information sources, monitor competitor changes, and have a briefing generated automatically every week. Your knowledge base goes from static to alive.
Final words
The investor friend from the dinner table tried it too later. He said to me:
"McKinsey used to need at least two weeks just to get started on industry research. But now, one person working alone with Workbuddy can get 80-90% of the information in an hour."
At the end of the day, this method isn't teaching you how to research an industry — it's teaching you how to turn learning an industry into a process, a tool, a switch.
(The data in the images is for reference only and does not represent actual data.)
I'm Xile. I'm using AI to build a one-person company and documenting the entire journey from 0 to 1.
If you also want to keep learning AI long-term and don't want to be left behind by the times, I've set up a learning group — just message me privately. It's for anyone who wants to learn AI, wants to start an AI side business, and doesn't want to be eliminated by the times.